Those of you who have taken to heart my posts on measuring your website’s success are measuring your website like crazy. You’re busy capturing data and examining the insights provided by your KPIs to understand exactly how your users are “voting” with their clicks. Right?
Now, when looking at a user’s click stream (or, path of click-to-click actions), you may start to have questions about what can be changed in order to improve your user’s likelihood of taking a desired action.
Testing is the next logical step. Of course, as with analyzing, you’ll have to choose the testing technology that’s right for your business. I recommend…
… Google Analytics "Content Experiments" for the same reasons that I recommended Google Analytics, in general, (because it’s free and relatively easy to use); however, it comes with the same privacy concerns, as well.
There are countless providers out there. Just pick one and get on with it! Testing can give you valuable insights that will spur change—much more rapidly than using nothing at all.
You can test a lot of different things, but it doesn’t necessarily mean you should. (Just because you can test 60 different button colors, doesn’t mean that’s the best use of your time.) Your testing should be guided by business intelligence and your experience.
Furthermore, keep in mind that just because a certain change “worked” for another company’s website, doesn’t mean it will work for yours. They may have entirely different objectives. For instance, it’s possible that when your competitor added required fields to be filled out on a form, it actually produced fewer leads…but resulted in more targeted ones. That decrease could be considered a successful action, if their desired outcome was to improve the quality of their leads.
Keep your eye on your prize; look at the things on your site that influence conversion the most—whether that means ringing up more sales or increasing the number of contact forms filled out by visitors. Then, start where the impact is greatest (which, for many websites, will be the homepage or another key entry page).
Although you can choose to test anything, we suggest that you start with something that will really “show me the money” (i.e. noticeably improve conversions). This may be altering the layout of an entire page or the design of a single element or different wording of headline text or the number of required fields in a form and so on. (An excellent tool for testing landing pages is LeadPages.)
You made certain assumptions when you created your website (e.g. “Visitors won’t want to buy something as soon as they land on the homepage.”). Now it’s time to test those assumptions. (Maybe, by the time new visitors get to your site, they’ve already done all of their research and are, in fact, ready to make a purchase.)
So, you need to employ the scientific method. You all remember that from junior high, don’t you? You have to come up with a hypothesis, figure out what the goal of your experiment is, and determine what “success” means for this particular experiment. You don’t have to know all the answers; you’re just comparing versions. The process goes like this:
- It’s relatively easy to create a hypothesis. It needs to include two important things: what you are going to do and what you expect the outcome to be.“If we add a ‘Buy It Now’ button to the top of our homepage, for a blue tennis racket that we’re currently running a promotion for, then sales of that particular tennis racket will increase.”
- Based on your hypothesis, you know what the goal is.“If the “Buy It Now” button does, in fact, increase sales of that blue tennis racket, then we will add a button to the homepage.”
- As with any scientific experiment, you need to have a control—something that you’re measuring against.“Our control will be our current homepage, exactly as it is, without any added buttons.”
- Be sure to compare apples to apples! Creating two dissimilar versions (i.e. one that looks like your current site with a “Buy It Now” button added, and one that has a tropical theme and music without a “Buy It Now” button), will give you results that will be difficult to interpret.“Sales of blue tennis rackets increased, but we’re not sure if that was due to the new button or the ukulele.”
- The last step is about determining what constitutes a successful outcome.“If we see a 10% increase in sales of the blue tennis racket, after adding a ‘Buy It Now’ button to the homepage, then we will consider the test to be a success.”
- After all of those decisions have been made, and you’ve walked through the entire scientific method, you can set up your test using the testing technology of your choice.
So, tell me: What assumptions have you made about your website (consciously or subconsciously) that you may need to test?
